In 0c77f7e9, the supported Octave version has
been changed from the Homebrew binary to the standalone application.

However, the runtime check about the supported Octave version was not updated,
and was still pointing to the Homebrew version number.

Note that MSYS2 is in the middle of the transition from xz to zstd for
compressing its archives, so we have to handle both formats.

Ref. #1727

Rather use a single vector as in non-block mode.

By the way, change the order of output arguments in static functions, to be
closer to the dynamic ones.

This cas was omitted in previous commit.

Rather use a single vector as in non-block mode.

Those cases are handled by dynare_solve.

— allow solve_algo=9 without block nor bytecode
— allow solve_algo≥9 with block without bytecode
— test solve_algo=9

Ref. #1724

This is necessary if we want our signatures to remain valid after the
expiration of our certificate.

For more details, see:
https://www.digicert.com/blog/best-practices-timestamping/

initval_file and hisvfal_file are now more flexible and have functionalities
similar to option datafile in estimation.

Closes: #1671

Bug introduced in previous commit.

— various refactoring and bugfixes in block decomposition
— small bugfix for PAC models

Closes: #1717

Ref. #1713

— Efficiency change. OCTAVE_VERSION is a builtin function in Octave
— Fix bug dseries#43
— Cosmetic changes
— Fixed subsasgn method (when calling merge)

Do not print the total computing time at the end of the driver evaluation.

missing_DiffuseKalmanSmootherH1_Z.m: fix bug with variable dimension

See merge request !1728

Fixes TaRB with Student t proposal, update manual, unit test

See merge request !1727

fix bug when no convergence and backward model

See merge request !1726

— Fix several bugs in computation of renormalized equations
— Various refactorings

— Block decomposition without bytecode: fix call to solve_one_boundary
  (Closes: #1717)
— Skip constant equations replacement only when there is an “mcp” tag (#1697)

— This is consistent with the fact that we hardcode the old MATLAB version
— This ensures better homogeneity across runners (in theory, it could be
  possible that different runners have different “matlab” symlinks)
— This allows easier customization in personal repositories